On 2 October 2017 at 14:22, Jules Richardson via cctech <cctech@classiccmp.org> wrote: > > Does anyone know why IDE/ATA even came about? I mean, why SCSI wasn't used?
Sure, yes. It was cheap. SCSI was expensive, and that was aside from any licensing issues. A working SCSI bus effectively means 2 smart devices, communicating over a defined _shared_ channel. ST-506 was simple, dumb and cheap... like the IBM PC. https://en.wikipedia.org/wiki/ST-506 As drive capacities grew, ESDI came along. https://en.wikipedia.org/wiki/Enhanced_Small_Disk_Interface I believe partly due to timing issues, some of the "smarts" of the controller were moved from the disk controller card onto the drive electronics -- but the cables were kept the same. (2 cables, 34 pin control cable, with 3 connectors, shared by up to 2 drives; plus 2× 20-pin data cables, one per drive.) Then most of the controller electronics were moved onto the drive, so that no "disk controller" was needed any more -- the drive contained the controller. Now the 2 cables were consolidated into a single 40-way cable, one end of which connected to the motherboard and the AT bus. (The 16-bit bus from the IBM PC-AT, so called to distinguish it from the 8-bit bus of the IBM PC.) The cable had 2 connectors for 2 drives, but they had to be jumpered to tell them which was master and which was slave, and not all combinations worked, not in the early days. IDE was mainly an x86 PC thing at first. Later, the 2nd-generation Acorn ARM machines had it, and the Commodore Amiga 1200 had an on-board interface for a 2.5" notebook-sized IDE drive. Later still Apple adopted it for its cheap home machines, the Performa PowerMacs. In the early ones, I think the hard disk was IDE and the CD drive was SCSI, and they still included a SCSI bus. Around 1994 or so, I ordered some machines from Elonex intended to run Windows NT 3.1. There was a substantial saving on ordering IDE models instead of SCSI-equipped ones. What I didn't know is that these had 540MB drives, and the IDE definition only allowed for 1024 cylinders, 63 sectors per track and 16 heads -- meaning 528MB max. These were not IDE drives, they were EIDE drives. Enhanced IDE. NT 3.1 didn't understand EIDE. It couldn't do Logical Block Addressing, only Cyl/Head/Track. So to NT, the last 12MB of these drives was all bad blocks. Returns and much argument followed. That got me a mention in Microscope magazine, which later got me an interview at Dennis Publishing and a job on PC Pro magazine. 21 years later, I'm a tech writer at SUSE in Prague. Funny how life turns out. Later, other limits came in at ~8GB, then at ~128GB, then a rather different one at 2TB. http://philipstorr.id.au/pcbook/book4/hdlimit.htm -- Liam Proven • Profile: https://about.me/liamproven Email: lpro...@cix.co.uk • Google Mail/Talk/Plus: lpro...@gmail.com Twitter/Facebook/Flickr: lproven • Skype/LinkedIn/AIM/Yahoo: liamproven UK: +44 7939-087884 • ČR/WhatsApp/Telegram/Signal: +420 702 829 053
